Conflicts of interest: None.
Immunocytochemistry suggests that the prevalence of a sub-type of beta-proteins determines the hardness in the epidermis of the hard-shelled turtle
Article first published online: 19 NOV 2013
© 2013 Wiley Periodicals, Inc.
Journal of Experimental Zoology Part B: Molecular and Developmental Evolution
Volume 322, Issue 1, pages 54–63, January 2014
How to Cite
2014. Immunocytochemistry suggests that the prevalence of a sub-type of beta-proteins determines the hardness in the epidermis of the hard-shelled turtle. J. Exp. Zool. (Mol. Dev. Evol.) 322B:54–63..
- Issue published online: 9 DEC 2013
- Article first published online: 19 NOV 2013
- Manuscript Accepted: 11 OCT 2013
- Manuscript Revised: 5 OCT 2013
- Manuscript Received: 6 JUL 2013
- Comparative Histolab
- University of Bologna. Grant Number: RFO 2010
The corneous layer of the epidermis in hard-shelled turtles largely derives from the accumulation of beta-proteins as indicated by microscopic, in situ hybridization, and immunocytochemical and Western blotting analysis. The expression of mRNAs of one of the most common type of beta-proteins shows higher expression in upper spinosus and pre-corneous keratinocytes of growing scutes. Two beta-proteins of 14–16 kDa, indicated as Tu2 and Tu17 and representing two subtypes of beta-proteins co-accumulate in the thick corneous layer of the epidermis in hard-shelled turtle. The two beta-proteins apparently mix in differentiating and mature corneocytes although Tu2 appears more prevalent than Tu17. The specific role of the different subtypes in the formation of the hard corneous material of the carapace and plastron is not clear. It is hypothesized that the relative amount of beta-proteins belonging to the two subclasses in relation to the alpha-keratin meshwork present in keratinocytes contributes to the formation of a variably resistant and inflexible corneous layer. Tu17 may have a more globular structure than Tu2 and is likely present in denser areas of the corneous layer containing also alpha-keratin. The increase of cysteine–glycine-rich beta-proteins in the matrix located among alpha-keratin filaments may allow the formation of a hard corneous material, probably through increase of cross-bridge formation and hydrophobicity. J. Exp. Zool. (Mol. Dev. Evol.) 322B: 54–63, 2014. © 2013 Wiley Periodicals, Inc.
The body plan of chelonians comprises the formation of a resistant shell composed of bony plates and horny scutes (Zangerl, 1969; Gilbert et al., 2001; Nagashima et al., 2012). The epidermis of the dorsal carapace and of the ventral plastron is similar, and in the resting phase it consists of a monolayered epidermis, few suprabasal keratinocytes, and a thick corneous layer (Alexander, 1970; Baden and Maderson, '70; Alibardi, 2002). However, during the periods of growth, the living part of the epidermis becomes thicker and several layers of fusiform and differentiating keratinocytes are produced from the basal layer to be incorporated into the corneous layer (Alibardi, 2005, 2006).
The differentiation of the epidermis in the shell (carapace and plastron) of turtles involves the production of hard and small proteins termed beta-keratins (Wyld and Brush, 1979, 1983; Holmer et al., 2001; Alibardi et al., 2004; Alibardi and Toni, 2005b). These small proteins however are not keratins but represent one of the main types of proteins that accumulate in the differentiating keratinocytes of the epidermis in turtles, and the other reptiles. Recent studies indicated that proteins formerly named beta-keratins represent instead (alpha-) keratin-associated beta-proteins (KAbetaPs) produced in differentiating keratinocytes of reptilian epidermis. Some of their genes and the complete sequences for 27 KAbetaPs have been reported in two species of turtle, the hard shelled turtle Pseudemys nelsoni and the soft-shelled turtle Apalone spinifera (Dalla Valle et al., 2009, 2013).
Based on their sequences and homology, two principal types of KAbeta-proteins have been found in the turtle P. nelsoni, Pn-Tu1–16 with similar amino acid identity, and only Pn-Tu17 with poor amino acid identity with the other 16 beta-proteins (Dalla Valle et al., 2009). The distribution of these proteins in the skin of turtles is not known. Recent molecular and immunocytochemical work (Alibardi, 2013; Dalla Valle et al., 2013) has shown that at least 10 KAbetaPs are expressed in the epidermis of the shell and other skin regions in the soft-shelled turtle A. spinifera, and that all these proteins possess higher identity to Pn-Tu17. The above studies therefore suggest that at least two subfamilies of KAbetaPs are present in turtles, but their chemical difference, characteristics and localization in the skin are unknown. Based on the sequence information for these proteins, two antibodies directed toward unique epitopes present in one member of the first subfamily were produced, indicated as Tu2 (Pn-Tu2), and another antibody directed against one member of the second subfamily Tu17 (Pn-Tu17). These two antibodies have been utilized in order to determine the localization of these proteins in the epidermis of the shell of freshwater, hard-shelled turtle P. nelsoni. The study aims to improve our understanding on the contribution of different beta-proteins for the formation of an epidermis with hard or soft characteristics, a process that influences the movement, swimming, and protection performances in turtles.
MATERIALS AND METHODS
Animals and Microscopic Methods
Three specimens of the freshwater, hard-shelled turtle P. nelsoni (Pn) were utilized for the study as previously indicated (Dalla Valle et al., 2009). Briefly, after sacrifice samples 2 × 2 mm2 large from the marginal scutes of the carapace and plastron were collected and fixed for 5–8 hr in 4% paraformaldehyde in 0.1 M phosphate buffer at pH 7.4, dehydrated in ethanol, and embedded in Bioacryl Resin at 0–4°C under UV light.
Using an ultramicrotome the tissues were sectioned at 2–3 µm thickness and the sections were stained with 1% toluidine blue for the histology examination. Other sections at 3–5 µm in thickness were utilized for the immunocytochemical reactions (see below). Three antibodies produced in rabbit were employed in the immunocytochemical study. One was a general antibody for detecting beta-proteins (beta-keratins, indicated as Beta-1 antibody, see Sawyer et al., 2000). The other two antibodies were epitope-specific and were directed to selected epitopes of Tu2 and Tu17 beta-proteins located at the C-terminus of the proteins (Fig. 1). One antibody, indicated as Pn-Tu2 (AN: AM765815), is directed against the epitope GLWGYGGYGRRYLGGRCGTC, while the other antibody, indicated as Pn-Tu17 (AN: FM 163397) is directed against the epitope YGKGYGRKCYSSRFGSCGPC. These epitopes are not present in (alpha-) keratins according to a BLAST search at http://blast.ncbi.nlm.nih.gov.
The immunization protocol and the purification of the antibody by affinity chromatography from rabbit sera using the selected peptides were carried out through a Biotechnology company (Davids Biotechnologie, Regensburg, Germany). The serum reactivity to the selected epitope was tested by ELISA, and showed a strong reactivity.
For immunofluorescence the sections were incubated 3 hr at room temperature with the primary antibodies at a 1: 100–200 dilution in Buffer (Tris 0.05 M at pH 7.6 containing 1% BSA). Sections were rinsed in buffer and incubated for 1 hr at room temperature with an anti-rabbit IgG secondary antibody conjugated to FITC diluted in Tris buffer at 1:80. After rinsing in the buffer, the antigen–antibody complex was detected with a fluorescence microscope. In control sections, the primary antibody was omitted from the incubating medium.
For TEM-immunogold, sections 60–90 nm thick were collected on Nickel grids using an ultramicrotome, and were incubated for 3 hr at room temperature in the primary antibodies diluted 1:100 in 0.05 M Tris–HCl buffer at pH 7.6, containing 1% Cold Water Fish Gelatin. In controls, the primary antibody was omitted in the overnight incubation. The sections were rinsed in buffer and incubated for 1 hr at room temperature with anti-rabbit Gold-conjugated secondary antibody (Sigma, St Louis, MI; Biocell, Cardiff, UK, 10 or 5 nm gold particles) for the ultrastructural localization. Grids were rinsed in buffer, stained for 5 min in 2% uranyl acetate and observed under the electron microscope Zeiss 10C/CR.
In Situ Hybridization
As representative of transcriptional expression of most beta-proteins in turtle epidermis (Tu1–16), the cRNA anti-sense and sense probes for the coding region of Tu2 beta-protein of P. nelsonii were utilized as previously described (Dalla Valle et al., 2009). Briefly, small pieces of the carapace were fixed in 4% paraformaldehyde, dehydrated, embedded in wax, and sectioned with a microtome. Sections 6–8 µm thick were incubated for the protocol of in situ hybridization using the anti-sense-cRNA probes for Tu2 coding region, and using sense-cRNA probes as control. The hybridizing medium contained 50% formamide, 4 × SSC (Standard Saline Citrate Solution), Tween-20 (1 µL/mL), tRNA (200 µg/mL), 0.1% Chaps, 0.5 mM EDTA, heparin (50 µg/mL), and 2% blocking reagent (Roche, Mannheim, Germany). For hybridization, an overnight incubation at 60°C in the hybridization buffer containing 0.5–1.5 ng/µL of digoxigenin-labeled probe was carried out. Hybridization was followed by successive washes at decreasing concentrations of SSC (2 × SSC, 0.5 × SSC, 0.2 × SSC, 0.1 × SSC, to increase stringency conditions) until the final phosphate-saline-Tween-buffer (PBT buffer). Sections were incubated for 2 hr at room temperature with anti-digoxigeninfluorescein Fab fragment antibodies diluted 1:15 in Tris Buffer (Roche, Mannheim, Germany). The labeling was detected using a fluorescence microscope. Other sections after hybridization were instead incubated with anti-digoxigenin alkaline phosphatase-conjugated antibody (Roche) diluted 1:500 in PBT buffer. Detection was performed with PBT buffer containing 4-nitroblue tetrazolium chloride (NBT) and 5-bromo-4-chloro-3-indolyl phosphate (BCIP) as substrates, as indicated by the manufacturer (Roche) in order to reveal the alcaline phosphatase activity as a purple-reddish stain. These sections were mounted in permanent medium and studied under a bright field optical microscope.
Electrophoretic Separation and Immunoblotting
For Western blotting some shed pieces of the superficial part of the stratum corneum of the carapace (Alibardi, 2005, 2006) were collected and washed several times with distilled water to clean them from dust. The scales were finely cut into tiny pieces (<1 mm large) before proteins extraction using the method of Sybert et al. (1985). Accordingly the scale fragments were homogenized in 8 M urea/50 mM Tris–HCl (pH 7.6)/0.1 M 2-mercaptoethanol/1 mM dithiothreithol/1 mM phenylmethylsulfonyl fluoride. This operation was repeated until most of the corneous material was no longer visible in the dense solution. The homogenate was left for 2 hr in water bath at about 70°C, and then at room temperature for an additional 2 hr before centrifugation. The insoluble material obtained after homogenization was removed by centrifugation at 10,000g for 10 min. Protein concentration was assayed by the Bradford method and, before electrophoresis analysis, the proteins were denatured by boiling the extracted solution in the Sample Buffer for 5 min and cooled.
Each electrophoretic lane was loaded with 35 µg of proteins and separated in 14.5% SDS–polyacrylamide gels (SDS–PAGE) according to Laemmli. Marker proteins were in the 6–70 kDa range of molecular weight marker for low molecular weight proteins (BioRad Laboratories Srl, Milan, Italy). For Western blotting, the proteins separated in SDS–PAGE were transferred to nitrocellulose paper. After Western blot, membranes were stained with Ponceau red to verify the protein transfer and incubated with primary antibodies directed against Tu2 or Tu17 antibodies (diluted 1:500 or 1:1,000). In controls, the primary antibody was omitted. After rinsing and fluorescent-conjugated anti-rabbit secondary antibody (1:1,000), the detection was performed using the enhanced chemiluminescence procedure developed by Amersham (ECL Plex Western Blotting System, GE, Healthcare, Waukesha, WI, USA).
General Biochemical Characteristics
The analysis of some representative beta-proteins (Tu5 and Tu8 as examples for the first sub-type, and of Tu17 for the second sub-type, see Fig. 2) using the Protparam Program at http://web.expasy.org/protparam/ showed that the content in cysteine, glycine, proline, serine, valine, and tyrosine, the most abundant and characteristic amino acids of beta-proteins (see Discussion section), is respectively 6–7%, 23–33%, 7–11%, 5–10%, 11–12%, 7–8% (Fig. 2). Therefore, there is no substantial difference apart from the position of some amino acids, especially of a central cysteine only in Tu17 (see cys–pro hinge in Fig. 1). The remaining part of the protein comprises a cysteine-rich N-region and a glycine-rich tail region after the central beta-pleated region where previous studies indicated the presence of anti-parallel beta-sheets (Alibardi et al., 2009). In the C-termini of all these beta-proteins, some cysteines are also located, and are included in the selected epitopes (Fig. 1). The prediction for these proteins of 14–17 kDa also indicate they are lightly basic and charged positively.
The use of the beta-1 antibody showed that only the stratum corneum is immunofluorescent in the monolayered epidermis (resting) indicating that beta-proteins are accumulated in corneocytes (Fig. 3A).
The in situ hybridization observations using the Tu2 anti-sense-probe showed expression only in growing (differentiating) areas of the carapace epidermis, when the epidermis becomes stratified. Using the anti-sense riboprobe the reddish staining was low to absent in lower epidermal layers, became intense in upper fusiform spinosus and pre-corneous keratinocytes, and disappeared in the fully cornified layers (Fig. 3B, C). The sense control showed much lower or no staining (Fig. 3D). Also the detection using anti-sense riboprobes and detection by fluorescence showed similar expression, although autofluorescence of the external part of the corneous layer was sometimes observed, as compared with controls (asterisks in Fig. 3E–G). While basal keratinocytes were not fluorescent, more intense fluorescent fusiform keratinocytes were seen in the upper spinosus and pre-corneous layers, and the specific fluorescence disappeared in the corneous layer (Fig. 3E, F). In sense controls no fluorescent keratinocytes were seen, and only the outer part of the corneous layer appeared sometimes fluorescent (Fig. 3G).
Immunoblotting and Immunocytochemistry
Immunolabeled bands were observed in the predicted range of molecular weight, at 15–16 kDa for Tu2 and at 14 and 16 kDa for Tu17 (1st and 2nd lanes in Fig. 4). In negative controls no bands were observed (third lane in Fig. 4).
The carapace comprised a columnar epithelium where few suprabasal cells were seen in resting phase while the thick stratum corneum measured 80–100 µm or more in some regions (Fig. 5A). The stratum corneum appeared broken upon sectioning revealing along the line of fracture the presence of thinner corneocytes than in the other, more compact regions of the stratum corneum. Along the thinner corneocytes is located the region of the stratum corneum forming the shedding line along which the outer part of the stratum corneum is shed and released as flakes once a year (Alibardi, 2005). Immunofluorescence using the Tu2 antibody revealed that the entire stratum corneum of the carapace and plastron was positive in resting phase while the living epidermis appeared immunonegative (Fig. 5B, C). Also the Tu17 antibody produced a positive immunofluorescence of the stratum corneum, although slightly weaker than using the Tu2 antibody (Fig. 5D, E). In numerous sections it also appeared that the splitting line was more immunofluorescent than the remaining parts of the corneous layer (Fig. 5D). In control sections no immunolabeling was seen or some weak non-specific fluorescence was observed in the outermost line of the corneous layer (Fig. 5F).
The ultrastructural localization of the Tu2 beta-protein showed that only the corneous layer of the resting epidermis was labeled while the basal keratinocytes were unlabeled (Fig. 6A). The labeling in the stratum corneum was uneven and some regions within the corneous layer were less intensely labeled than others. Keratin bundles, including those converging on the desmosomes connecting keratinocytes to the corneous layer, were unlabeled, including those embedded in the corneous material of corneocytes (inset of Fig. 6A). The distribution of gold particles was relatively homogenous over mature corneocytes although the prevalent paler areas were more labeled than darker areas (Fig. 6B, and inset at higher magnification).
In various regions within the thick corneous layer, corneocytes showed some areas along the plasma membrane still separated from other corneocytes, and connected by desmosomal remnants (Fig. 6B, inset). No other region of the epidermis as well as non-epidermal tissues were labeled indicating the unique presence of this epitope in the corneous material of the carapace or plastron. The gold labeling of the corneous layer was not homogenously distributed but showed regions with higher labeling in comparison to other regions. Also higher magnification images of paler areas of the corneous layer showed the higher labeling than in darker areas, although the latter represented a minor part of the corneous material (Fig. 7A, B). In the hinge regions among scutes, the labeling decreases and sparse gold particles were seen among the filaments of alpha-keratin (Fig. 7C).
Also the Tu17 immunogold labeling was absent in the living keratinocytes forming the basal layer or in those connected to the corneous layer (Fig. 8A). The labeling appeared unevenly distributed over the condensed material of the stratum corneum and was generally more diluted with respect to the Tu2 immunoreactivity (Fig. 8B). In some areas with no apparent patterned distribution, gold particles were also densely distributed as previously observed with the Tu2 antibody (Fig. 8C). A diffuse distribution of gold particles was also seen in the thinner cells of the likely shedding line that showed a similar labeling present in other corneocytes (Fig. 8D). The immunolabeled areas for Tu17 included those more electron-dense and the labeling for Tu7 appeared less intense and selective than the labeling for Tu2. No labeling was seen in other tissues and in controls (data not shown).
Localization and Characteristics of the Two Types of Beta-Proteins
The expression observations and the restricted immunolocalization of beta-proteins in the stratum corneum confirm previous autoradiographic and expression studies indicating that newly produced beta-proteins only occur in growing epidermis and not in resting epidermis where no cells are added to the corneous layer of the carapace or plastron (Alibardi et al., 2004; Alibardi, 2005, 2006; Alibardi and Toni, 2005a; Dalla Valle et al., 2009, 2013).
The present Western blot results indicate that the two proteins are found in the expected range of 14–17 kDa, supporting the following immunolocalization observations. The two subtypes have a similar composition (Fig. 2) and therefore their main difference appears in the location of the cysteine–proline hinge region present after the beta-sheet region, the more conserved region of beta-proteins (see the cysteine–proline hinge region in Fig. 1; Alibardi and Toni, 2007; Alibardi et al., 2009). Both proline, an amino acid involved in folding protein chains, and valine, which is involved in building the central beta-sheet region of these proteins (Fraser et al., 1972; Fraser and Parry, 2011), are similar in percentage in both subtypes of turtle beta-proteins.
The much higher content in cysteine of turtle beta-proteins in comparison to alpha-keratins, 6–7% versus 0.6–1.4%, and of glycine, 23–33% versus 4–9% (data deduced using the Protparam Program at http://web.expasy.org/protparam/ on the sequences published in Dalla Valle et al., 2013), suggests that the smaller beta-proteins act as matrix molecules for the alpha-keratin meshwork (Fig. 9). The cysteines, especially those localized at the extremities of beta-proteins (Fig. 1) are likely involved in the cross-linking and possible hardening of the corneous material (Fraser et al., 1972) in the corneous layers of hard-shelled turtles. Conversely, it is likely that a prevalence of alpha-keratins over beta-proteins in the corneous layer of the soft-shelled turtle (Dalla Valle et al., 2013), due to the low amount of potentially reactive cysteines, is one of the factors that determine a lower cross-linking and therefore the formation of a softer and pliable corneous material in these turtles (Fig. 9).
The presence of the cysteine–proline hinge in the Tu17 like protein (Fig. 1), a dipeptide that is missing in the Tu2 protein, may determine some differences in the space conformation of these two types of beta-proteins. While Tu2 (representing 16 beta-proteins, Tu1–16, see Dalla Valle et al., 2009) may give rise to more linear molecules, the Tu17 type may instead form a more globular protein due to the presence of the inner cysteine available to form an intramolecular disulfide bond with other cysteines. In case of formation of this internal bond, the latter might determine the compaction of the three-dimensional shape of the amino acid chain, so inducing a more globular conformation of this specific protein with respect to the Tu2 protein. Besides, the potential presence of an intra-molecular –S–S– bond within the beta-region of these proteins may influence the polymerization process (see Discussion below).
The effect of mixing these two different sub-types of proteins and their interaction with alpha-keratins on the texture of the resulting corneous material (Fig. 9) remains to be evaluated. The indications of the present immunolocalization tentatively suggest that Tu2-like beta-proteins are localized in the more electron-pale regions of the corneous material where alpha-keratins appear absent or limited (Alibardi and Toni, 2006). Conversely, the presence of Tu17-like beta-proteins also in denser regions of the corneous layer may suggest that the latter beta-proteins interact with alpha-keratins and other proteins present in these regions. In fact, previous studies have shown that during the differentiation of new corneocytes in hard-shelled turtles, some clumps of dense material of unknown composition are mixed with paler beta-keratin pachets (Alibardi et al., 2004; Alibardi, 2006). This also occurs in the soft-shelled turtle where the incorporated denser material also contains beta-proteins (Alibardi, 2013).
Although light microscopy has indicated that the splitting line contains more Tu17 than the remaining corneocytes, the ultrastructural study has not conclusively demonstrated that this is the case, and more studies are needed on this point. Despite a previous study showed that the splitting region also contains alpha-keratin (Alibardi, 2005), it is still unknown whether a possible association of Tu17 with specific alpha-keratins occurs in this special region of the stratum corneum in turtles.
Hypothesis on the Origin of Scales With Different Rigidity
The specific properties of the corneous material present in different regions of the skin, in the shell versus the limb, neck, and tail (Baden and Maderson, 1970) may depend from the ratio between alpha- and beta-proteins and from the ratio between Tu2-like and Tu17-like beta-proteins (Dalla Valle et al., 2013). The hypothesis is presented in Figure 9 but this issue can only be resolved by analyzing the three dimensional conformation and interaction among these proteins and the interactions with other proteins.
In hard-shelled turtles, the Tu2 subfamily is most represented while the Tu17 subfamily is less represented. The first subtype of beta-proteins can polymerize into filaments and give rise to the hard corneous material typical of hard-shelled turtles (A–A6 in Fig. 9). Conversely, in the soft-shelled turtle A. spinifera, all the beta-proteins so far isolated cluster with Tu17 in P. nelsonii (Figs. 3 and 7 in Dalla Valle et al., 2013), and possess the cys–pro hinge region. This characteristic may interfere with the formation of the beta-protein filament since no polymerization can occur among this subtype of beta-proteins (B–B6 in Fig. 9). The presence of higher amount of alpha-keratins and of the Tu17 like beta-proteins is correlated with the softness of the corneous layer in the epidermis. Therefore, it is hypothesized that when there is more Tu17 than Tu2 in addition to a prevalence of alpha-keratin, the corneous layer is softer, like in the hinge regions and softer scales of hard-shelled turtles and in the soft-shelled turtle. This hypothesis however should be experimentally verified.
The study was largely self-supported (Comparative Histolab) and in part with a University of Bologna RFO 2010 Grant.
- 1970. Comparison of alpha and beta keratin in reptiles. Z Zellforsch 110:153–165. .
- 2002. Immunocytochemical observations on the cornification of soft and hard epidermis in the turtle Chrysemys picta. Zoology 105:31–44. .
- 2005. Proliferation in the epidermis of chelonians and growth of the horny scutes. J Morphol 265:52–69. .
- 2006. Ultrastructural and immunocytochemical observations on the process of horny growth in chelonian shell. Acta Histochem 108:149–162. .
- 2013. Ultrastructural immunolocalization of alpha-keratins and associated beta-proteins (beta-keratins) suggests a new interpretation on the process of hard and soft cornification in turtle epidermis. Micron 52–53:8–15. .
- 2005a. Immunolocalization and characterization of beta-keratins in growing epidermis of scutes and claws of chelonians. Tissue Cell 38:53–63. , .
- 2005b. Skin structure and cornification proteins in the soft-shelled turtle Trionyx spiniferus. Zoology 109:182–195. , .
- 2007. Immunological characterization of a newly developed antibody for localization of a beta-keratin in turtle epidermis. J Exp Zool 308B:200–208. , .
- 2004. Differentiation of the epidermis in turtle: an immunocytochemical, autoradiographic and electrophoretic analysis. Acta Histochem 106:379–395. , , .
- 2009. Evolution of hard proteins in the sauropsid integument in relation to the cornification of skin derivatives in amniotes. J Anat 214:560–586. , , , .
- 1970. Morphological and biophysical identification of fibrous proteins in the amniote epidermis. J Exp Zool 174:225–232. , .
- 2009. Beta-keratins of turtle shell comprise small glycine–proline–tyrosine rich proteins similar to those of crocodilians and birds. J Anat 214:284–300. , , , .
- 2013. Molecular characterization of keratins and associated beta-proteins in soft-shelled and hard-shelled turtles produced during the process of epidermal differentiation. J Exp Zool 320B:428–441. , , , , .
- 2011. The structural basis of the filament-matrix texture in the avian/reptilian group of hard β-keratins. J Struct Biol 173:391–405. , .
- 1972. Keratins: their composition, structure and biosynthesis. Springfield, IL: Charles C. Thomas. , , .
- 2001. Morphogenesis of the turtle shell: the development of a novel structure in tetrapod evolution. Evol Dev 3:47–58. , , , .
- 2001. Soluble scute proteins of healthy and ill desert tortoises (Gopherus agassizii). Am J Vet Res 62:104–110. , , , et al.
- 2012. Body plan of turtles: an anatomical, developmental and evolutionary perspective. Anat Sci Int 87:1–13. , , , , , .
- 2000. The expression of beta (β) keratins in the epidermal appendages of reptiles and birds. Am Zool 40:530–539. , , , et al.
- 1979. The molecular heterogeneity and diversity of reptilian keratins. J Mol Evol 12:331–347. , .
- 1983. Keratin diversity in the reptilian epidermis. J Exp Zool 225:387–396. , .
- 1985. Ichthyosis vulgaris: identification of a defect in synthesis of filaggrin correlated with an absence of keratohyaline granules. J Invest Dermatol 84:191–194. , , .
- 1969. The turtle shell. In: Gans C, editor. Biology of the reptilia. vol 1. London: Academic Press. p 311–339. .